Several people have got this disk full error recently. When SQLite runs a VACUUM, it writes a couple of copies of client.db, one in your temp dir (%appdata%/local/temp for me), and another next door to client.db, at client.db-journal. For people synced to my PTR, client.db is about 3GB, so if you have less than 3GB free on your system drive or hydrus install dir, you get this error. I have written a little text file in install_dir/db about it, and I will now update the popup error message now we better understand what causes this.
If you cannot free up a few gigs on your system drive, you can leave vacuuming disabled. Vacuuming is basically db defrag–it is not very important, but nice to run every now and then.
I am thinking of breaking up client.db into smaller dbs in the coming month or so, which will alleviate this problem (and several others, hopefully!).
If you client is running slow, please check out this page:
http://hydrusnetwork.github.io/hydrus/help/reducing_lag.html
Now I think about it, if vacuum is already disabled but your client still isn't closing cleanly, then something else might be going on as well. Please check install_dir/logs/client.db and scroll to the bottom. Are there any database errors, likely after a 'shutting down gui' line?